draft modelled estimates for the spread and health impact of … · draft modelled estimates for...
TRANSCRIPT
1
Draft
Modelled Estimates for the Spread and Health Impact of Covid-19 in New
Zealand: Revised Preliminary Report for the NZ Ministry of Health
27 February 2020
Dr Lucy Telfar Barnard, Prof Nick Wilson, Dr Amanda Kvalsvig, Prof Michael Baker (HEIRU,
University of Otago Wellington; contact: [email protected])
Acknowledgements: We thank our Australian colleagues for their valuable work in providing
modelling data on numbers of infected cases for different scenarios. Nevertheless, the additional
analyses and interpretation in this document are those of the above named authors alone and do not
necessarily represent the views of our Australian colleagues or the New Zealand Ministry of Health.
Key messages
• The future spread and health impact of this new disease are highly uncertain at present as
there is limited evidence about the full spectrum of transmissibility, infection severity and
impact of control measures. This Report therefore can only represent potential future
scenarios rather than predictions.
• If substantially uncontrolled spread of disease occurs in NZ during 2020, our “plan for”
scenario based on the modelling work by our Australian colleagues sees 65% of the
population (3,230,000) contracting a Covid-19 infection, though only 34% of the population
(1,680,000) would probably be symptomatic (typically mild disease). However, these could
well be over-estimates if the disease is less transmissible in NZ than China (at least for
January 2020 in China), if the summer and spring conditions in NZ slow the spread of disease
(noting that other coronavirus infection is typically a winter phenomenon), and if control
measures are more effective than we anticipate. Indeed, our “plan for” scenario is somewhat
towards “worse case” and the most likely outcome for NZ is probably less severe.
• Health service demand is likely to be extremely high in our “plan for” scenario as we
estimate that 336,000 people are likely to require hospitalisation, with a worst day demand of:
7,300 new patients, and 1,450 to 1,700 new intensive care unit (ICU) patients. This high
demand risks overloading the hospital sector (given very limited surge capacity in NZ
hospitals and ICUs at present).
• The peak week in our “plan for” scenario is 21 weeks after virus introduction (for
uncontrolled spread), but successful control measures would postpone the date of the peak by
some weeks to months.
• We estimate likely deaths to be between 12,600 and 33,600 people in our “plan for” scenario.
This range can be considered relatively severe compared with our scenarios where lower
transmissibility is modelled. It is also much more severe than the results of a case study
where we take disease parameters from the largest outbreak outside of China ie, the Diamond
Princess cruise ship outbreak and use these to extrapolate to all of NZ (see Appendix 2). Our
“plan for” mortality estimates could also be over-estimates if treatment options improve and
if a vaccine arrives later in 2020.
• As in previous epidemics and pandemics of infectious respiratory agents, severe disease
burden is likely to fall unequally on Māori, Pacific peoples, and the elderly (see Appendix 3).
2
Background
This report provides numerical estimates and analytical overlay of potential New Zealand Covid-19
scenarios, based on modelling of infected case numbers from McVernon et al and using a NZ
population of 4.96 million (unpublished report provided 21 February 2020).
Method
Graphics of models from McVernon et al were overlaid onto a grid. Data were extracted visually,
and graphed in Excel to extract estimates of total infection and peak day numbers.
Assumptions
This report provides estimates for three potential epidemic scenarios for two levels of infectivity
(R0). All scenarios assume:
• Pre-symptomatic transmission as indicated in a Japanese study1 (though the extent of this is
still very uncertain).
• 48% of infections being asymptomatic (Japanese cruise ship data2);
• 20% of symptomatic infections requiring hospitalisation (ie, not in the “mild” category3). The first, “plan for” infection scenario uses an R0 of 2.2. This was the estimated reproduction
number reported in Li et al 2020,4 which we judge had the least methodological issues of measures to
date.
We also provide numbers for an R0 of 1.5. While an R0 of 1.5 is lower than many early published
estimates, it is consistent with the 2009 H1N1 influenza pandemic for NZ (mid-range out of 3
estimates5) and is more consistent with previous pandemics than values of 2.0 or higher (see
Appendix 2).
For each scenario, we provide:
• Two alternative estimates for ICU demand, one based on a very large Chinese study
indicating that 4.7% of cases are “critical”;3 and the other based on our estimate that 4% of
symptomatic cases are likely to require ICU care.6
• Three alternative estimates for deaths, the first based on the WHO estimated case fatality risk
(CFR) of 2% of symptomatic cases; the second based on our estimate that 34.1% of ICU
admissions end in death6; and the third based on the 21 February 2020 CFR for cases outside
China of 0.75%.7 (The latter was prior to evidence suggestive of uncontrolled disease spread
being plausible in Iran and Italy).
Interpretation and commentary
The large degree of uncertainty in the epidemiology of Covid-19 prevents us from providing narrow
estimates for potential Covid-19 infections.
Our “plan for” scenario (which can be considered to be towards the “worse case” end of the outcome
spectrum) sees 3,230,000 people, or 65% of the population, becoming infected with Covid-19,
though we only expect 34% of the population (1,680,000) to experience symptoms. Nonetheless,
even with these infection numbers, current estimates of the severity of the disease suggest health
services are likely to be over-burdened, with 336,000 people being likely to require hospitalisation,
including a peak demand day of 7,300 additional patients progressing to a stage requiring
hospitalisation. Between 67,000 and 79,000 people are likely to need ICU treatment, with peak day
demand of 1,450 to 1,700 additional ICU patients.
3
Our “plan for” scenario estimates total deaths between 12,600 and 33,600 people. These mortality
estimates could be over-estimates if disease spread is slower (as per R0 = 1.5), and if control
measures are particularly successful; or they could still be under-estimates given hospitals and ICU
overload will probably increase the case fatality risk. However, we note that this range is
substantially higher than our estimate for the NZ mortality burden based on disease parameters for
the largest outbreak outside of China ie, the Diamond Princess cruise ship outbreak (see Appendix
2). Extrapolating from that outbreak, we estimated 6,230 or 6,510 deaths in NZ if the same disease
parameters applied in NZ as applied to this cruise ship. We expect transmission on the cruise ship
was likely to have been very high, but the passengers also likely had a much higher average age than
the NZ population. On the other hand, this cruise ship outbreak has been largely truncated via people
being moved off the ship.
International data suggest that deaths will also typically be in those with co-morbidities: this topic
can be addressed in a future report.
Figure 1: First 5 weeks of epidemic growth, showing all infections regardless of the degree of
symptoms, for R0=1.5, with 48 hrs pre-symptomatic transmission, by effectiveness of control
measures (from McVernon et al)
4
Figure 2: Epidemic curves showing ALL infections regardless of the degree of symptoms, for
R0=1.5 and R0=2.2, with 48 hrs pre-symptomatic transmission, by effectiveness of control
measures (from McVernon et al)
5
Table 1: Estimates for Covid-19 spread in NZ, 48 hrs pre-symptomatic transmission
R0 1.5* 2.2
Infection control effectiveness
25% 0% 50% 25% (“plan for” scenario)
0%
Peak day Week 47 Week 33 Week 27 Week 21 Week 17
All infections
Total 1,490,000 2,370,000 2,630,000 3,230,000 3,560,000
Proportion of NZ population (%)
30% 48% 53% 65% 72%
Worst day 13,500 31,000 42,000 70,000 96,000
Symptomatic cases
Total 773,000 1,230,000 1,370,000 1,680,000 1,850,000
Worst day 7,000 16,000 22,000 36,000 50,000
Severe cases likely to require hospitalisation
Total 155,000 246,000 273,000 336,000 370,000
Worst day 1,400 3,200 4,300 7,300 10,000
Cases likely to require ICU
Total WHO (4.7%) Total UoO (4%)
36,000 31,000
58,000 49,000
64,000 55,000
79,000 67,000
87,000 74,000
Peak WHO (4.7%) Peak UoO (4%)
330 280
800 640
1,000 860
1,700 1,450
2,350 2,000
Deaths (CFR)
Total WHO (2%) Total (ICU x 34%) Total (0.75)%
15,500 10,500 5,800
24,600 16,800 9,200
27,300 18,630 10,200
33,600 23,000 12,600
37,000 25,000 14,000
Peak WHO (2%) Peak (ICU x 34%) Total (O.75%)
140 100 50
320 220 120
430 290 160
730 500 270
1,000 680 380
* Figures for R0=1.5 at 50% infection control are not precisely ascertainable from currently available graphical output from the modelling work by our Australian colleagues, more detail can potentially be made available in a subsequent report.
6
Limitations of this analysis
1. There is a high degree of uncertainty around all aspects of Covid-19 epidemiology. The R0
may well be lower than our 2.2 “plan for” scenario value or it may be higher. Similarly, the
CFR could be overestimated (due to missing mild cases in the denominator) or
underestimated (due to the lag period for deaths to occur in ICU or from missed deaths).
2. The models do not adjust for potential seasonal effects, as these are complex. Disease spread
is likely to be initially dampened if Covid-19 arrives in the warmer part of autumn or if case
numbers remain low until late spring. Conversely, winter conditions would increase disease
spread. The timing in the modelling for disease spread in NZ, did not consider potential
super-spreading events. If these occurred it could truncate the timeline to the epidemic peak.
3. This analysis does not consider potential internal travel restrictions to prevent spread to
another island or region (eg, when there is uncontrolled spread in the North Island – then
protective sequestration could be applied to the South Island or of regions with limited road
access such as the West Coast). Eg, the pandemic plan for Iceland identifies where road
blocks would be set up to stop pandemic spread within the main island.
4. This analysis also does not consider protective sequestration of institutions such as retirement
villages (ie, this was done for some institutions in the 1918 pandemic with some success).
This intervention could impact on reducing the mortality burden since older people appear to
have higher CFRs.
5. While this analysis uses such estimates as “25%” and “50%” control – it is possible that even
more intense control is plausible eg, as per the above mentioned travel restrictions, protective
sequestration and various other social distancing options. Also mass media campaigns by the
government to promote staying at home when sick and to enhance hand and respiratory
hygiene, could also reduce transmission. Indeed, the recent decline in new cases in China
suggests that it is quite plausible that control measures could be more effective than the
“50%” used in the modelling (albeit with some of the control measures in China being
relatively severe).
6. Estimates for numbers of deaths assume people can receive necessary hospital and ICU
treatment, but as health services will potentially be over-burdened, the CFR is likely to
increase.
7. The models do not account for the potential discovery or development of more effective
treatments. This innovation could reduce the number of cases needing hospitalisation and
intensive care, and consequently reduce deaths.
7
Appendix 1: Case study of extrapolating from a large cruise ship outbreak of
Covid-19 to NZ (if this pandemic reaches NZ and becomes uncontrolled)
An advantage of considering the outbreak on the Diamond Princess cruise ship is that it provides a
clearer idea of the disease parameters and denominator populations – given the widespread testing of
the crew and passengers. While attempts at quarantine were made on the cruise ship, this does not
appear to have been fully effective and conditions on the ship are far more crowded than typical
community settings (and indeed the ship crew were in dormitories and continued to have contact
with passengers). Also the age distribution of passengers was probably older than the New Zealand
population – and so this may increase the risk of more severe outcomes, including death. On the
other hand, this cruise ship outbreak has been largely truncated via people being moved off the ship.
Also the calculations below make no allowance for how overloaded hospitals and ICUs may not
function normally when the pandemic peaks in a country like NZ, with the case fatality risk (CFR)
being likely to increase in overloaded settings.
Table A1: A case study using parameters from a large cruise ship outbreak of Covid-19 and
extrapolating to the NZ situation (if the pandemic reaches NZ and becomes uncontrolled)
Characteristic
Diamond Princess
(cruise ship) data
and estimates
Extrapolations
to NZ Comment
People with Covid-
19 infection
(including some
asymptomatic
infection)
634 people on 20
February 2020,8 so
this was 17.1% of
the 3711 (crew and
passengers)2
initially on the ship
847,000 We used this proportion of cases
(17.1%) for the NZ population (4.96
million population). Of note, however,
is that some testing on the cruise ship
may still be proceeding and so this
proportion may increase with time.
Another complexity in terms of final
interpretation of the outbreak size, is
that the denominator has started to
change with evacuations from the
ship. Of note is that those passengers
with lab-confirmed COVID-19 were
disembarked and transferred to an
isolation ward at healthcare facilities.
Severity at level
requiring ICU
2.1% 17,700 On 18 February when 454 cases were
reported on the cruise ship: “The
health ministry said 19 of the infected
people were in serious condition, with
some of them in intensive-care
units.”9 So for this calculation we
assumed half of these 19 people were
in ICU.
Estimated deaths
(extrapolating from
ICU numbers)
5* 6050 We used the 34.1% from the typical
case fatality in ICUs for severe
respiratory conditions10 (and for extra
context and workings see: Wilson et
al6).
8
Characteristic
Diamond Princess
(cruise ship) data
and estimates
Extrapolations
to NZ Comment
Estimated deaths
using an out-of-
China CFR
5* 6320 Out-of-China CFR as per a WHO
report on 20 February (8/1073 =
0.75%.)7
* Indeed, three people from the cruise ship have actually died as detailed in a 24 February WHO Situation Report (where
the number of cases/people testing positive had reached 695).
9
Appendix 2: Estimating the basic reproduction number (R0) for COVID-19 in
New Zealand
The basic reproduction number R0 represents the average number of secondary infections per case in
a fully susceptible population. Future projections of clinical burden and impact are highly sensitive
to the value of R0 selected for modelling as this value drives the proportion of the population
infected. Estimating the value of R0 is challenging, however, because estimates of R0 are themselves
sensitive to both measurement error and environmental conditions, particularly during an evolving
pandemic. Measures of R0 in a given population will strongly reflect the sociodemographic
characteristics, public health interventions, climate, and geography of that population.5 11
Because there have been relatively few cases outside China (and none to date in New Zealand), most
of the currently available estimates are based on cases occurring in China during the early stages of
the epidemic. Estimates of R0 calculated in this way vary from 2 to 5.12
R0 is consistently overestimated in the early phases of a pandemic, for a range of reasons.13 Taking
this bias into account together with contextual differences, it is highly likely that the true R0 in New
Zealand would be substantially lower than estimates from Chinese studies. Reasons for the
difference include:
• Higher levels of crowdedness in Chinese cities compared with New Zealand
• Initial overestimation of R0 e.g. because of changes in reporting rates.12
• Greater population susceptibility to infection from existing compromised respiratory health
due to poor air quality from air pollution and high smoking rates in China.
Table A2 shows estimates of R0 in the New Zealand population during respiratory virus pandemics.
Values of R0 have been consistently in the range 1 to 2, with one slightly higher early estimate.
Table A2. Estimates of the reproduction number in New Zealand during previous respiratory
virus outbreaks.
Estimated R0 (95% CI) Pandemic Source Comment
1.96 (1.80 – 2.15) 2009 H1N1 Nishiura et al., 200914 Preliminary estimate
1.25 (1.07 – 1.47) 2009 H1N1 Roberts et al., 201115 Updated estimate
1.55 (1.16 – 1.86) 2009 H1N1 Paine et al., 2010 16 Peak value
1.34 (1.27 – 1.38) 2009 H1N1 Opatowski et al., 201111 Confirmed cases
1.2 – 1.8 1918 pandemic Wilson et al., 201217 In community settings
These New Zealand estimates are also consistent with global estimates of R0 from previous
pandemics; retrospective estimates of R0 from systematic reviews and meta-analyses include:
• 1.2 to 1.8 in eight Southern Hemisphere countries for 2009 H1N1 pandemic influenza11
• 1.65 (IQR 1.53 – 1.70) in the 1957 influenza pandemic5
• 1.80 (IQR 1.47 – 2.27) in the 1918 influenza pandemic.5
During its early stages, Severe Acute Respiratory Syndrome (SARS) was estimated in modelling to
have an R0 of 2.2 – 3.6, but data from the first 205 probable cases in Singapore indicated an R0 < 1
10
from approximately the third week onwards.18 Mercer et al list several similar examples of this
trend.13
Summary
Currently available estimates of R0 are uncertain because the transmission characteristics of this
novel virus are still somewhat unclear. The “plan for” value of R0 = 2.2 indicates the current
consensus around scenarios for planning. However, estimates generated by data from China are
likely to introduce substantial overestimation of impacts in the New Zealand population. For this
reason, scenarios based on a lower value of R0 are also presented for comparison purposes. The value
of R0 = 1.5 included in the models presented here is consistent with a) the lower limit of estimates
from China, b) estimated R0 in other New Zealand pandemics, and c) global estimates from other
pandemics that were generated after the earliest stages of the event.
11
Appendix 3: Potential age and ethnic distribution of Covid-19 health impacts
(assumed uncontrolled spread occurs in NZ).
Key messages
• The numbers presented here represent potential future scenarios rather than predictions
and will be strongly influenced by the transmissibility and severity of the Covid-19
pandemic when it reaches New Zealand.
• If Covid-19 follows the same patterns as previous pandemics, there may be a relatively high and
heavily unequal hospitalisation and mortality burden on Māori and Pacific populations.
• Elderly are particularly at risk, and Māori and Pacific elderly even more so, and from younger
ages.
• Nevertheless, these poor outcomes are modifiable ie, if the entry of Covid-19 to NZ is delayed, if
there are successful interventions that substantially reduce transmission in Māori and Pacific
communities (eg, intensive hand and respiratory hygiene messages and social distancing
interventions such as reducing public gatherings and closing schools) or even more radical
measures are used such as protective sequestration of communities.
The numbers in this Appendix are based on the “plan for” scenario of R0=2.2, 48hrs pre-
symptomatic transmission, with 25% effective infection control measures. We use the Chinese data
(CCDC) 2% case fatality risk3 as our basis for age-specific deaths, and we treat the age distribution
of cases in that report as the age distribution of hospitalisations.
Ethnic distribution
We assume ethnic inequalities in mortality will be similar to those in the 2009 Influenza A(H1N1)
pandemic, ie, that the risk of hospitalisation was 5 times higher for Māori and 7 times higher for
Pacific peoples,19 while the risk of death was 2.6 times higher for Māori (95%CI: 1.3 – 5.3) and 4.6
times higher for Pacific peoples (95%CI: 2.0 – 7.2) than for NZ European/Other.20
The estimated ethnic distribution of the hospitalisation and mortality burden under the “plan for”
scenario is shown in Table A3-1. Numbers do not include any age-standardisation.
Table A3-1. Estimated ethnic distribution of Covid-19 hospitalisations and deaths, “plan for”
scenario.
Ethnic group Hospitalisations Deaths
n Pop % n Pop%
Māori 128,230 16.4% 9,240 1.2%
Pacific 92,190 23.1% 8,360 2.1%
NZ European/Other 115,580 3.3% 16,000 0.5%
Total 336,000 6.8% 33,600 0.7%
Age distribution
Our estimates of the age distribution of deaths (Table A3-2) are based on CCDC reports of case
fatality risk by age, and the Chinese population age structure estimated from the CIA World
Factbook, adjusted to the New Zealand population and the projected number of deaths.
12
Table A3-2. Estimated age distribution of Covid-19 mortality, “plan for” scenario.
Age group (years) Hospitalisations Deaths
n Pop % n Pop %
0-9 3,132 0.5% 0 0.0%
10-19 4,173 0.7% 19 0.0%
20-29 21,193 3.0% 102 0.0%
30-39 47,807 7.4% 282 0.0%
40-49 43,689 7.0% 482 0.1%
50-59 65,690 10.4% 2,122 0.3%
60-69 63,542 12.2% 5,689 1.1%
70-79 41,181 11.9% 8,155 2.3%
80+ 45,593 25.2% 16,750 9.3%
Total 336,000 6.8% 33,600 0.68%
Age and ethnic distribution
The age and ethnic disaggregated figures which follow are based on multiple interpolations and
extrapolations. As these numbers are a number of steps removed from their bases, they can only be
very rough estimates. Numbers for Pacific peoples over the age of 60 were particularly unstable. The
key message from these numbers is that the mortality burden is likely to fall particularly heavily on
older (aged ~60+) Māori and Pacific peoples
NZ European/Other
Estimated NZ European/Other deaths under the “plan for” scenario are shown in Table A3-3.
Table A3-3. Estimated NZ European/Other age distribution of Covid-19 hospitalisations and
mortality, “plan for” scenario.
Age group (years) Hospitalisations Deaths
n Pop %* n Pop %
0-9 1,021 0.2% 0 0.0%
10-19 1,353 0.3% 8 0.0%
20-29 6,666 1.4% 42 0.0%
30-39 13,426 3.5% 104 0.0%
40-49 14,137 3.3% 205 0.0%
50-59 23,265 4.9% 987 0.2%
60-69 23,382 5.7% 2,748 0.7%
70-79 16,539 5.6% 4,300 1.4%
80+ 15,788 11.8% 7,615 5.7%
Total 115,580 3.3% 16,009 0.46%
* Percentages are row percentages (with the denominator being the population in that age-group).
Māori and Pacific peoples
There are difficulties extrapolating mortality distributions from the Chinese example to the Māori
and Pacific populations, who not only have a very different population age structure, but also have
different life expectancies. Comparing population mortality rates, in ordinary circumstances, the
mortality risk of a Māori person aged 60 years, for example, is roughly equivalent to the mortality
risk of a NZ European person aged 70 years.
13
Since Covid-19 mortality appears to heavily reflect age-related vulnerability, we have calibrated
Māori and Pacific age-specific mortalities to NZ European/Other age-specific mortality when
calculating estimated ethnic-group specific age-related mortality. Calibrating age bands reduces the
starting age of the upper “80+ years” age band to 74 years for Māori and 73 years for Pacific
peoples.
Following this calibration, estimates for age-related mortality for Māori and Pacific peoples under
the “plan for” scenario are shown in Tables A3-4 and A3-5. In Table A3-5 the 60+ and 73+ age
bands were combined, as low population numbers in the top band resulted in an unstable estimate.
Table A3-4. Estimated Māori age distribution of Covid-19 hospitalisations and mortality, “plan
for” scenario.
Age group (years) Hospitalisations Deaths
n Pop % n Pop %
0-9 2,475 1.5% 0 0.0%
10-19 3,030 1.9% 10 0.0%
20-29 17,305 13.5% 68 0.1%
30-39 17,945 20.8% 144 0.2%
40-49 25,404 30.7% 598 0.7%
50-59 27,412 36.0% 1,788 2.3%
60-73 21,472 35.0% 3,098 5.0%
74+ 13,186 74.3% 3,529 19.9%
Total 128,230 16.4% 9,236 1.2%
Table A3-5. Estimated Pacific age distribution of Covid-19 hospitalisations and mortality,
“plan for” scenario.
Age group (years) Hospitalisations Deaths
n Pop % n Pop %
0-9 1,949 2.1% 0 0.00%
10-19 2,266 2.8% 10 0.0%
20-29 14,393 20.2% 76 0.0%
30-39 14,760 30.5% 158 0.3%
40-49 18,330 45.2% 575 1.4%
50-59 17,935 53.0% 1,559 4.6%
60+ 20,099 59.8% 4,739 14.1%
Total 92,190 23.1% 8,355 1.2%
14
References
1. Nishiura H, Linton NM, Akhmetzhanov AR. Serial interval of novel coronavirus (2019-nCoV)
infections. medRxiv 2020:2020.02.03.20019497. doi:
https://doi.org/10.1101/2020.02.03.20019497
2. National Institute of Infectious Diseases. Field Briefing: Diamond Princess COVID-19 Cases.
NIID (Government of Japan). 19 February 2020. https://www.niid.go.jp/niid/en/2019-ncov-
e/9407-covid-dp-fe-01.html.
3. The Novel Coronavirus Pneumonia Emergency Response Epidemiology Team. The
Epidemiological Characteristics of an Outbreak of 2019 Novel Coronavirus Diseases
(COVID-19) — China, 2020. China CDC Weekly 2020.
http://weekly.chinacdc.cn/en/article/id/e53946e2-c6c4-41e9-9a9b-fea8db1a8f51.
4. Li Q, Guan X, Wu P, et al. Early Transmission Dynamics in Wuhan, China, of Novel
Coronavirus-Infected Pneumonia. New England Journal of Medicine 2020;29:29. doi:
https://dx.doi.org/10.1056/NEJMoa2001316
5. Biggerstaff M, Cauchemez S, Reed C, et al. Estimates of the reproduction number for seasonal,
pandemic, and zoonotic influenza: a systematic review of the literature. BMC Infectious
Diseases 2014;14(1):480. doi: 10.1186/1471-2334-14-480
6. Wilson N, Kvalsvig A, Telfar Barnard L, et al. Estimating the Case Fatality Risk of COVID-19
using Cases from Outside China. MedRxiv (pre-print archive) 2020;(15 February). doi:
https://doi.org/10.1101/2020.02.15.20023499
https://www.medrxiv.org/content/10.1101/2020.02.15.20023499v1.
7. World Health Organization. Coronavirus disease 2019 (COVID-19) Situation Report – 31.
2020;(20 February). https://www.who.int/docs/default-source/coronaviruse/situation-
reports/20200220-sitrep-31-covid-19.pdf?sfvrsn=dfd11d24_2.
8. Belam M, Quinn B, Rourke A. Coronavirus: cruise ship accounts for more than half of cases
outside China – as it happened. The Guardian 2020;(20 February).
https://www.theguardian.com/world/live/2020/feb/20/coronavirus-live-updates-diamond-
princess-cruise-ship-japan-deaths-latest-news-china-infections.
9. Newstalk ZB/NZ Herald. 99 more cases confirmed on Diamond Princess cruise ship. Newstalk
ZB/NZ Herald 2020;(18 February). https://www.newstalkzb.co.nz/on-air/early-
edition/audio/david-murdoch-on-coronavirus-99-more-cases-confirmed-on-diamond-
princess-cruise-ship/.
10. Briel M, Meade M, Mercat A, et al. Higher vs lower positive end-expiratory pressure in patients
with acute lung injury and acute respiratory distress syndrome: systematic review and meta-
analysis. JAMA 2010;303(9):865-73. doi: 10.1001/jama.2010.218 [published Online First:
2010/03/04]
11. Opatowski L, Fraser C, Griffin J, et al. Transmission characteristics of the 2009 H1N1 influenza
pandemic: comparison of 8 Southern hemisphere countries. PLoS Pathog
2011;7(9):e1002225-e25. doi: 10.1371/journal.ppat.1002225 [published Online First:
2011/09/01]
12. Zhao S, Lin Q, Ran J, et al. Preliminary estimation of the basic reproduction number of novel
coronavirus (2019-nCoV) in China, from 2019 to 2020: A data-driven analysis in the early
phase of the outbreak. Int J Infect Dis 2020;92:214-17. doi:
https://doi.org/10.1016/j.ijid.2020.01.050
13. Mercer GN, Glass K, Becker NG. Effective reproduction numbers are commonly overestimated
early in a disease outbreak. Statistics in Medicine 2011;30(9):984-94. doi: 10.1002/sim.4174
14. Nishiura H, Wilson N, Baker MG. Estimating the reproduction number of the novel influenza A
virus (H1N1) in a Southern Hemisphere setting: preliminary estimate in New Zealand. The
New Zealand Medical Journal (Online) 2009;122(1299)
15
15. Roberts MG, Nishiura H. Early estimation of the reproduction number in the presence of
imported cases: pandemic influenza H1N1-2009 in New Zealand. PLoS One 2011;6(5)
16. Paine S, Mercer G, Kelly P, et al. Transmissibility of 2009 pandemic influenza A (H1N1) in New
Zealand: effective reproduction number and influence of age, ethnicity and importations.
Eurosurveillance 2010;15(24):19591.
17. Wilson N, Summers JA, Baker MG. The 2009 influenza pandemic: a review of the strengths and
weaknesses of the health sector response in New Zealand. Clinical Correspondence 2012
18. Lipsitch M, Cohen T, Cooper B, et al. Transmission Dynamics and Control of Severe Acute
Respiratory Syndrome. Sci 2003;300(5627):1966-70. doi: 10.1126/science.1086616
19. Verrall A, Norton K, Rooker S, et al. Hospitalizations for pandemic (H1N1) 2009 among Maori
and Pacific Islanders, New Zealand. Emerg Infect Dis 2010;16(1):100-02. doi:
10.3201/eid1601.090994
20. Wilson N, Barnard LT, Summers JA, et al. Differential mortality rates by ethnicity in 3 influenza
pandemics over a century, New Zealand. Emerg Infect Dis 2012;18(1):71-77. doi:
https://doi.org/10.3201/eid1801.110035